clean clean clean (#1626)

nucliadb_vectors/src/data_point/mod.rs

@@ -29,7 +29,7 @@ use std::collections::{HashMap, HashSet};
 use std::time::SystemTime;
 use std::{fs, io, path};
 
-use data_store::Slot;
+use data_store::Interpreter;
 use disk_hnsw::DiskHnsw;
 use io::{BufWriter, Write};
 use memmap2::Mmap;
@@ -236,7 +236,7 @@ impl<'a, Dlog: DeleteLog> Retriever<'a, Dlog> {
             nodes,
             delete_log,
             similarity,
-            no_nodes: data_store::number_of_written_slots(nodes),
+            no_nodes: data_store::stored_elements(nodes),
             min_score,
         }
     }
@@ -344,15 +344,7 @@ impl Elem {
     }
 }
 
-impl data_store::Data for Elem {
-    fn serialized_len(&self) -> usize {
-        Node::serialized_len(
-            &self.key,
-            &self.vector,
-            &self.labels.0,
-            self.metadata.as_ref(),
-        )
-    }
+impl data_store::IntoBuffer for Elem {
     fn serialize_into<W: io::Write>(self, w: W) -> io::Result<()> {
         Node::serialize_into(
             w,
@@ -441,7 +433,7 @@ impl AsRef<DataPoint> for DataPoint {
 
 impl DataPoint {
     pub fn stored_len(&self) -> Option<u64> {
-        if data_store::number_of_written_slots(&self.nodes) == 0 {
+        if data_store::stored_elements(&self.nodes) == 0 {
             return None;
         }
         let node = data_store::get_value(Node, &self.nodes, 0);
@@ -454,11 +446,18 @@ impl DataPoint {
         self.journal
     }
     pub fn get_keys<Dlog: DeleteLog>(&self, delete_log: &Dlog) -> Vec<String> {
-        data_store::get_keys(Node, &self.nodes)
-            .filter(|k| !delete_log.is_deleted(k))
-            .map(String::from_utf8_lossy)
-            .map(|s| s.to_string())
-            .collect()
+        let node_storage = &self.nodes;
+        let length = data_store::stored_elements(node_storage);
+        let data_iterator = (0..length).map(|i| data_store::get_value(Node, node_storage, i));
+        let mut keys = Vec::new();
+        for data in data_iterator {
+            if !delete_log.is_deleted(data) {
+                let raw_key = Node.get_key(data);
+                let string_key = String::from_utf8_lossy(raw_key).to_string();
+                keys.push(string_key);
+            }
+        }
+        keys
     }
 
     #[allow(clippy::too_many_arguments)]
@@ -481,7 +480,7 @@ impl DataPoint {
             min_score,
         );
 
-        let no_nodes = data_store::number_of_written_slots(&self.nodes);
+        let no_nodes = data_store::stored_elements(&self.nodes);
 
         let filter = FormulaFilter::new(
             filter,
@@ -542,7 +541,7 @@ impl DataPoint {
             .collect();
         data_store::merge(&mut nodes, &node_producers)?;
         let nodes = unsafe { Mmap::map(&nodes)? };
-        let no_nodes = data_store::number_of_written_slots(&nodes);
+        let no_nodes = data_store::stored_elements(&nodes);
 
         // Creating the FSTs with the new nodes
         let (label_index, key_index) = if channel == Channel::EXPERIMENTAL {
@@ -651,7 +650,7 @@ impl DataPoint {
         root_dir: &path::Path,
         nodes: &[u8],
     ) -> VectorR<(Option<LabelIndex>, Option<KeyIndex>)> {
-        let no_nodes = data_store::number_of_written_slots(nodes);
+        let no_nodes = data_store::stored_elements(nodes);
 
         // building the KeyIndex and LabelIndex FSTs
         let fst_dir = root_dir.join(file_names::FST);
@@ -731,7 +730,7 @@ impl DataPoint {
         // Nodes are stored on disk and mmaped.
         data_store::create_key_value(&mut nodesf, elems)?;
         let nodes = unsafe { Mmap::map(&nodesf)? };
-        let no_nodes = data_store::number_of_written_slots(&nodes);
+        let no_nodes = data_store::stored_elements(&nodes);
 
         // Creating the FSTs
         let (label_index, key_index) = if channel == Channel::EXPERIMENTAL {

nucliadb_vectors/src/data_point/node.rs

@@ -20,7 +20,7 @@
 
 use std::io;
 
-use crate::data_types::data_store::Slot;
+use crate::data_types::data_store::Interpreter;
 use crate::data_types::trie;
 use crate::data_types::usize_utils::*;
 
@@ -163,11 +163,7 @@ impl Node {
         trie::has_word(&x[xlabel_ptr..], label)
     }
 }
-impl Slot for Node {
-    fn cmp_keys(&self, x: &[u8], key: &[u8]) -> std::cmp::Ordering {
-        let xkey = self.get_key(x);
-        xkey.cmp(key)
-    }
+impl Interpreter for Node {
     fn get_key<'a>(&self, x: &'a [u8]) -> &'a [u8] {
         Self::key(x)
     }
@@ -269,7 +265,6 @@ mod tests {
         assert_eq!(Node::vector(&buf[node2..]), vector2);
         assert!(Node.keep_in_merge(&buf[node1..]));
         assert!(Node.keep_in_merge(&buf[node2..]));
-        assert_eq!(Node.cmp_slot(&buf[node1..], &buf[node2..]), key1.cmp(key2));
         assert_eq!(Node::metadata(&buf[node1..]), metadata1);
         assert_eq!(Node::metadata(&buf[node2..]), metadata2);
         assert_eq!(